Usage

svg_heatmap can be used as a drop-in replacement for seaborn.heatmap, with the exception of a few missing features:

• center
• annot
  • fmt
• linewidths, linecolor
• square
• mask

Please note that this does not use any internal seaborn functions, neither use matplotlib.pyplot.pcolormesh for the color mesh. It will therefore look slightly different than a corresponding seaborn plot.

Imports

import numpy as np; np.random.seed(0)
import seaborn as sns; sns.set()
from matplotlib import pyplot as plt
from svg_heatmap import heatmap
from ipywidgets import HTML
import sys
from io import BytesIO
import binascii

Comparing plots

Helper Function

def compare_plots(data, svg_kwargs={}, sns_kwargs={}, **kwargs):
    svg_plot = heatmap(data, **svg_kwargs ,**kwargs)
    
    fig=plt.figure()
    sns.heatmap(data, **sns_kwargs, **kwargs)
    plt.tight_layout()
    with BytesIO() as buf:
        fig.canvas.print_png(buf)
        png_data = binascii.b2a_base64(buf.getvalue()).decode()
        
    png_html = '<img src="data:image/png;base64,{}">'
    sns_png_plot= png_html.format(png_data)
    
    with BytesIO() as buf:
        plt.savefig(buf, format='svg')
        sns_svg_plot = buf.getvalue().decode()
        
    plt.close()
    
    svg_size, sns_png_size, sns_svg_size = [str(round(sys.getsizeof(plot) / 1024, 1)) + 'kB' 
                                        for plot in (svg_plot, sns_png_plot, sns_svg_plot)]
    return HTML('svg {}<br>'.format(svg_size) + svg_plot +'<br>sns svg {}<br>'.format(sns_svg_size) + sns_svg_plot
               +'<br>sns png {}<br>'.format(sns_png_size) + sns_png_plot)

ndarray data

compare_plots(np.random.rand(10, 12), cmap='viridis')

DataFrame data

flights = sns.load_dataset("flights")
flights = flights.pivot("month", "year", "passengers")
compare_plots(flights, cmap='magma')

No cbar

compare_plots(flights, cmap='magma', cbar=False)

log scaling

data_w_outliers = np.random.rand(10, 12)
data_w_outliers[2:3,3:6] += 5*data_w_outliers.max()

from matplotlib.colors import LogNorm
compare_plots(data_w_outliers, cmap='magma', svg_kwargs=dict(log_scaling=True), 
              sns_kwargs=dict(norm=LogNorm(vmin=data_w_outliers.min(), vmax=data_w_outliers.max())))

SVG vs png cbar

cbars are generated with pyplot and embedded in the root SVG. This can be done by either base64 encoding the png output or using the SVG output.

However, the SVG output is significantly larger since it uses <path>s instead of <text>.

svg_cbar_plot = heatmap(np.random.rand(10, 12), svg_cbar=True)
png_cbar_plot = heatmap(np.random.rand(10, 12), svg_cbar=False)

svg_cbar_size, png_cbar_size = [str(round(sys.getsizeof(plot) / 1024, 1)) + 'kB' 
                                for plot in (svg_cbar_plot, png_cbar_plot)]

HTML('svg cbar {}<br>'.format(svg_cbar_size) + svg_cbar_plot +
     '<br>png cbar {}<br>'.format(png_cbar_size) + png_cbar_plot)